What is kodwai?

kodwai is a platform where developers solve real coding challenges on their own machine using their own AI coding agent such as Claude Code, Cursor, or Codex. Instead of a browser sandbox, it runs through a CLI that downloads the problem, starter files, and tests, then scores your session out of 100. Scoring emphasizes how you direct the agent (Direction), what shipped (Outcome), and how far you beat a solo AI (Lift).

How to use kodwai?

1. Install Node.js 20+ and git, then run npx @kodwai/cli@latest challenge <slug> to pick a challenge; the CLI signs you in via browser and asks which agent you will use. 2. The CLI downloads PROBLEM.md, starter files and tests, sets up a git repo, and starts the timer. 3. Solve the challenge in your own editor with your own agent, planning, prompting, testing and pushing back. 4. Run npx @kodwai/cli@latest submit to send your code, git history, test runs and agent transcript for scoring.

Core features of kodwai

Real coding challenges solved on your own machine with your own AI agent (Claude Code, Cursor, Codex)
CLI-driven workflow: start a challenge and submit with two commands
AI Collaboration Score out of 100 across Direction, Outcome, and Lift axes
Detailed rubric with dimensions like Spec Precision, Verification Rigor, Decomposition, Recovery, Intent Fidelity, Engagement
Hidden traps and key moments surfaced after scoring
Free for developers, with profile and leaderboard
Session replay comparing hands-off vs hands-on runs